nrel variability analysis for the western interconnect ... · nrel variability analysis for the...

NREL is a national laboratory of the U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, operated by the Alliance for Sustainable Energy, LLC.

NREL Variability Analysis for the Western Interconnect

WECC WebEx

Brendan Kirby, Jack King, Michael Milligan

National Renewable Energy Laboratory

July 29, 2011

NREL/PR-5500-52430

Composite photo created by NREL

National Renewable Energy Laboratory 2 Innovation for Our Energy Future

• Study Data• Methodology• Analysis• Allocation of reserve requirements

Contents


• Investigate the effects of several Energy Imbalance Markets implementations in the Western Interconnect (WI):– Operating reserves;– Ramping demand;– Alternative Schedu.ing

• Analysis based on aggregation of variability and uncertainty:– Uses available load, wind and solar data.

• Calculate the reserve requirements:– Estimating within the hour and hourly requirements

based on historical data.• Compared to NREL draft report, this analysis has

approximately 1/3 VG.

Overview


Study Data


Model Area

WALC

BPAPACW

EPE

CHPD

GCPD

DOPD

TPWR

SCL

TEP

SRP

AZPS

NEVPTID

SMUD

PNM

PSCO

SPP

PACE

WACM

NWE

IPC

PGE

AVA

BCTC

PSE

IID

• Based on TEPPC PC 0.• Areas of the WI not

already in a market structure (CAISO, Alberta).

• LADWP not included.• Analysis at BAA

granularity.• Generation only BAAs

not considered.• 28 load and generation

BAAs retained – The study ‘Footprint.’

• Regions are:– Columbia Grid – Orange;– NTTG – Blue;– West Connect – White.


• 2020 loads provided by WECC;• Load shape is based on 2006 load;• 193,700 MW in full Western Interconnect (non-

coincident).• 116,700 MW in analysis footprint (non-

coincident);• One hour load provided, 10-minute synthesized.

Load Data


• NREL WWSIS wind dataset for 2006;• 10-minute resolution;• Sites identified by TEPPC 2020 PC 0;• 29,085 MW in Western Interconnect;• Approximately 8% of WI 2020 load;• 18,272-MW nameplate modeled in analysis –

footprint.

Wind Data


• NREL WECC Solar dataset for 2006;• 10-minute resolution;• Sites identified by TEPPC 2020 PC 0;• 14,300 MW in WI;• Approximately 3% of WI 2020 load;• 4,568 MW nameplate in analysis footprint.

Solar Data


Reserve Calculations


• Developed for the EWITS study;*• Statistical approach based on 10-minute time

series wind, solar and load;• Method that can estimate adequate reserves to

cover the short term and hour-ahead forecast error based on historical data;

• Predicts requirements based on current hour load and wind, solar production;

• Statistically combine with load regulation requirements;

• Provide 8760 vector of requirements for the production simulations.

*For in-depth discussion see section 5 of the EWITS final report: http://www.nrel.gov/wind/systemsintegration/pdfs/2010/ewits_final_report.pdf

Reserve Calculations

http://www.nrel.gov/wind/systemsintegration/pdfs/2010/ewits_final_report.pdf


• Regulation – Fast changes:– Due to variability and short-term forecast errors;– Faster than re-dispatch period;– AGC resources required.

• Spinning – larger, slower, less frequent variations:– Due to longer term forecast errors;– AGC not required;– 10-minute response.

• Non-spinning/supplemental:– Large, infrequent, slow moving events such as

unforecasted ramps;– 30-minute response.

Reserve Definitions


Short-Term Forecast Error - Regulation

• Based on persistence forecas;t

• Wind data is 10 minute, 10-minute delay for forecast;

• Forecast error is calculated as difference from actual to forecast.


Short-term Calculation

• Measure as standard deviation of the forecast error;

• Forecast error varies with production level;• Empirical expected error as a function of production

is quadratic:– Low variability at low and high wind (cut-in and rated);– High variability in mid range at steep part of power curve;– Solar follows same pattern, more or less.

• Predicts the expected variability for the hour based on the intra-hour statistics, current production;

• Assumes fast dispatch, 10 minutes in this case:– Implication is that economic movement happen at 10-

minute updates.


Calculation of Short-term Forecast Error

• Calculate and sort error by production level;

• Divide production into deciles;

• Calculate error sigma in each decile;

• Blue line is calculated from data;

• Red line is curve fit;• Equation of the curve

shown below.


Calculating the Regulation Requirement


Hour-ahead Forecast Errors

• Repeat the short term forecast procedure with hour-ahead forecasting;

• Again, same procedure for wind and solar;• Load following not included.


Spin and Non-spin Calculation


Verification of Reserve Coverage

• Shows coverage of VG intra-hour movements;

• Actual ramp data from footprint EIM;

• Red line – Actual average reserve calcs from footprint EIM;

• Probability lines are z% of all ramps at x minutes are less than y MW.


Reserves Provided to E3 for Phase 2 Study

• NREL provided reserve calculations to E3 for phase 2 study;

• Slight variation in calculation of regulation component –load not included in E3 Flex;

• Does not include contingency reserves;• Areas are slightly different from each other.

NREL E3 Response Notes

Regulation Flex AGCE3 flex reserve does not contain load

Spin Spin <10 minutesNon-spin Supplemental <30 minutes Terminology change


Analysis


Cases

• Complete analysis includes:– Footprint EIM

• All 28 BAAs participating. – ‘Regional’ EIM implementations

• Columbia Grid;• Northern Tier Transmission Group;• WestConnect.

– EIM implementations without BPA and/or WAPA participation.

– Comparison to the Business-As-Usual (BAU) case –No EIM in place;

– Selected results.


Footprint EIM Regulation

• Detail of the individual regulation components;

• Bars are average values, whiskers are min and max;

• 3 regional results in smaller reductions compared to footprint.

Reduction over BAUFootprint Regional

MW%

Reduct. MW%

Reduct.Load Reg 211 18% 178 16%Wind Reg 372 53% 231 33%Solar Reg 49 21% 46 19%Total Reg 570 36% 419 26%


Reserve Details for Footprint and Regional EIM

• Up to 42% (2000 MW) reduction in total reserve requirement for footprint EIM;

• 26% for regional EIMs.

Reduction over BAUFootprint Regional

MW % Reduct. MW % Reduct.Total Reg 570 36% 419 26%Spin 481 45% 271 25%Non-spin 963 45% 542 25%Total 2014 42% 1233 26%


Ramp Demand Reduction – Footprint EIM

• Based on hourly ramps;• Shows ramp reduction

from EIM over BAU;• Duration plot shows

hours per year for reduction level;

• Reduction in net ramp is greater than 1000 MW for 251 hours per year and averages about 260 MW.


Reserve Reduction Duration – Footprint EIM

• Average reductions:– 590-MW Regulation;– 2092-MW Total Reserves.


Regional EIM – Columbia Grid (CG)

• Average reserves for a CG EIM;

• BPA dominates the wind so saving less than other regions;

• 8070 MW Wind.

Reserve ReductionMW % Reduc.

Load Reg 37 14%Wind Reg 39 15%Solar Reg 0 0%Total Reg 85 21%Spin 38 12%Non-spin 76 12%Total 199 15%


Regional EIM – NTTG

• Average reserves;• Very small solar;• Wind dominates

savings;• ~4200 MW wind

and 20 MW solar.




Regional EIM – WestConnect (WC)

• Average reserves;• Substantial saving

from wind and solar;• ~5700 MW Wind and

4550 MW solar.




Conclusions

• Method for calculating additional reserve requirements due to wind and solar production;

• EIM results in substantial reduction in reserves requirements and ramping demand;

• Reduced participation reduces benefits for all but reduces the benefits to non-participants the most;

• Full participation leads to maximum benefit across the Western Interconnection, up to 42% of total reserve requirement;

• Regional EIM implementations have smaller but substantial benefits.

nrel variability analysis for the western interconnect ... · nrel variability analysis for the...

Documents